Bias in journalism is rarely intentional — it is usually structural: which sources are included vs. omitted, which framings are presented as neutral vs. partisan, which communities are centred vs. marginalised. AI agents trained to analyse these structural patterns can flag potential bias before a story publishes, supplementing the human editor's perspective.
The Bias Detection Prompt
After completing a draft: "Analyse this article for potential journalistic bias. Specifically: (1) source balance — are all significant perspectives represented? Which perspectives are absent? (2) Framing — does the article present any contested position as the neutral default? (3) Language — are emotionally charged terms applied asymmetrically to different groups or positions? (4) Proportionality — does the space given to different positions reflect their actual weight in expert and public opinion? Do not simply endorse the article — your job is to flag concerns." The output typically identifies 2–5 specific concerns for editorial review.